Quantization made by Richard Erkhov.

Power-WizardLM-2-13b - GGUF

Model creator: https://huggingface.co/refine-ai/
Original model: https://huggingface.co/refine-ai/Power-WizardLM-2-13b/

Name	Quant method	Size
Power-WizardLM-2-13b.Q2_K.gguf	Q2_K	4.33GB
Power-WizardLM-2-13b.IQ3_XS.gguf	IQ3_XS	4.81GB
Power-WizardLM-2-13b.IQ3_S.gguf	IQ3_S	5.07GB
Power-WizardLM-2-13b.Q3_K_S.gguf	Q3_K_S	5.04GB
Power-WizardLM-2-13b.IQ3_M.gguf	IQ3_M	5.24GB
Power-WizardLM-2-13b.Q3_K.gguf	Q3_K	5.62GB
Power-WizardLM-2-13b.Q3_K_M.gguf	Q3_K_M	5.62GB
Power-WizardLM-2-13b.Q3_K_L.gguf	Q3_K_L	6.11GB
Power-WizardLM-2-13b.IQ4_XS.gguf	IQ4_XS	6.3GB
Power-WizardLM-2-13b.Q4_0.gguf	Q4_0	6.57GB
Power-WizardLM-2-13b.IQ4_NL.gguf	IQ4_NL	6.64GB
Power-WizardLM-2-13b.Q4_K_S.gguf	Q4_K_S	6.62GB
Power-WizardLM-2-13b.Q4_K.gguf	Q4_K	6.99GB
Power-WizardLM-2-13b.Q4_K_M.gguf	Q4_K_M	6.99GB
Power-WizardLM-2-13b.Q4_1.gguf	Q4_1	7.29GB
Power-WizardLM-2-13b.Q5_0.gguf	Q5_0	8.01GB
Power-WizardLM-2-13b.Q5_K_S.gguf	Q5_K_S	8.01GB
Power-WizardLM-2-13b.Q5_K.gguf	Q5_K	8.22GB
Power-WizardLM-2-13b.Q5_K_M.gguf	Q5_K_M	8.22GB
Power-WizardLM-2-13b.Q5_1.gguf	Q5_1	8.73GB
Power-WizardLM-2-13b.Q6_K.gguf	Q6_K	9.53GB
Power-WizardLM-2-13b.Q8_0.gguf	Q8_0	12.35GB

Original model description:

tags: - merge - mergekit - lazymergekit - WizardLM - WizardLM2 license: mit pipeline_tag: text-generation

Power-WizardLM-2-13b

Power-WizardLM-2-13b is a merge of the following models using LazyMergekit:

🧩 Configuration

slices:
- sources:
  - layer_range: [0, 8]
    model: dreamgen/WizardLM-2-7B
- sources:
  - layer_range: [4, 12]
    model: dreamgen/WizardLM-2-7B
- sources:
  - layer_range: [8, 16]
    model: dreamgen/WizardLM-2-7B
- sources:
  - layer_range: [12, 20]
    model: dreamgen/WizardLM-2-7B
- sources:
  - layer_range: [16, 24]
    model: dreamgen/WizardLM-2-7B
- sources:
  - layer_range: [20, 28]
    model: dreamgen/WizardLM-2-7B
- sources:
  - layer_range: [24, 32]
    model: dreamgen/WizardLM-2-7B
merge_method: passthrough
dtype: float16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "KingNish/Power-WizardLM-2-13b"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])